Li Jin
@li
2 replies
1 recast
13 reactions
:omer
@omer
Your point about data collection is spot on. Another related area with great potential is synthetic data generation: a decentralized network of data generation nodes scales much more efficiently compared to a data training network etc. because you can run subtasks in parallel
1 reply
0 recast
0 reaction
:omer
@omer
People contribute to an open knowledge hub -> Community annotation gets the knowledge ready for retrieval -> Use a decentralized vectordb for retrieval -> Data generation nodes retrieve from the knowledge hub + web, and the results are combined into a new dataset Each step is powered by crypto
0 reply
0 recast
0 reaction
